Abstract
Altered retinal neurovascular coupling may contribute to the development and progression of diabetic retinopathy (DR) but remains highly challenging to measure due to limited resolution and field of view of the existing functional hyperemia imaging. Here, we present a novel modality of functional OCT angiography (fOCTA) that allows a 3D imaging of retinal functional hyperemia across the entire vascular tree with single-capillary resolution. In fOCTA, functional hyperemia was evoked by a flicker light stimulation, recorded by a synchronized time-lapse OCTA (i.e., 4D), and extracted precisely from each capillary segment (space) and stimulation period (time) in the OCTA time series. The high-resolution fOCTA revealed that the retinal capillaries, particularly the intermediate capillary plexus, exhibited apparent hyperemic response in normal mice, and significant functional hyperemia loss (P < 0.001) at an early stage of DR with few overt signs of retinopathy and visible restoration after aminoguanidine treatment (P < 0.05). Retinal capillary functional hyperemia has strong potential to provide sensitive biomarkers of early DR, and retinal fOCTA would provide new insights into the pathophysiology, screening and treatment of early DR.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Diabetic retinopathy (DR), the most common ocular complication of diabetes mellitus (DM), is the leading cause of severe vision loss in the working-age population [1,2]. Early detection and timely intervention are critical for preserving sight before the occurrence of irreversible damage to the retina [3]. Retinal neurovascular coupling (NVC), a phenomenon also termed functional hyperemia, is a key mechanism that regulates blood flow to meet the demands of transient neural activity [4] and plays a crucial role in the development and progression of DR [5]. Alterations in retinal NVC are believed to be an early predictive biomarker for DR [6,7]. Measurement of functional hyperemia has been used in studies of DR and fundamentally relies on the analysis of vascular dynamic responses to retinal neuronal stimulation [8,9].
Flicker light stimulation (FLS) is widely used to elicit functional hyperemia because it can effectively evoke neural activation in the inner retina [9–11]. FLS-evoked arteriolar and venular dilation can be measured as retinal NVC function with a dynamic vessel analyzer (DVA), i.e., a modified fundus camera that presents FLS and records videos of fundus images [12,13]. However, functional hyperemia assessment that is limited to retinal arterioles and venules may not provide information on NVC function of specific layers and regional NVC dysfunction in DR [6]. In addition, retinal capillary functional hyperemia may play a more pronounced role in blood flow regulation under neural activation [14]. Until recently, retinal capillary functional hyperemia has remained inaccessible due to the limited spatial resolution of fundus cameras in DVA [13]. Although high-resolution confocal imaging has been used to investigate the stimulus-evoked retinal capillary dynamics [10], this approach has several limitations: 1) the need to inject a contrast agent, 2) a limited field of view (FOV) of only a few hundred micrometers, 3) limited imaging depth with visible light, and 4) interference in FLS by visible light from the imaging channel. These limitations may restrict the utility of high-resolution confocal imaging in clinical practice for frequent use and longitudinal observation.
Optical coherence tomography angiography (OCTA) has enabled three-dimensional (3D), label-free visualization of retinal microvasculature down to the capillary level [15–17] and has wide applications in clinical routine in ophthalmology [18]. By combining OCTA recording and light stimulation, retinal functional hyperemia has been measured either in an FOV of a single cross section [19] or on a level of the plexus layer [11,20]. In this report, we demonstrate that functional OCTA (fOCTA) extends retinal functional hyperemia imaging to a 3D FOV and single-capillary resolution. We show that fOCTA provides local estimates of FLS-evoked functional hyperemia across the entire retinal vascular tree in both normal and DM mice. We also investigate the heterogeneities of the functional hyperemia in different retinal vascular compartments during the development and treatment of early-stage DR.
2. Methods
2.1 Animal preparation
Male C57BL/6J mice obtained from the Zhejiang Medical Science Institute were used in this study. These mice were randomly assigned to streptozotocin-induced or control groups. In the streptozotocin group, DM was induced at 7 wk of age by intraperitoneal injection of streptozotocin (55 mg/kg body weight in 10 mmol/L sodium citrate buffer) on five successive days. The blood glucose concentration of tail vein blood was measured at 2 days after the last injection and monitored every half month using a portable blood glucose meter (OneTouch Ultra, Johnson & Johnson, USA). Only the mice with glucose levels consistently elevated above 300 mg/dL were recognized as DM mice (Table 1). The retina of the DM group (N = 15) was imaged with fOCTA at 4 wk after DM induction (age 12 wk). After the fOCTA imaging, 10 mice were injected aminoguanidine (AG) intravenously (100 mg/kg body weight), and the AG-treated mice were imaged within 4 h after AG injection. The other 5 mice of DM group were euthanized for immunohistochemistry analysis. The age-matched control group (N = 10) underwent retinal fOCTA imaging at age 12 wk.
All experiments were conducted in a dark room with ambient light blocked out. Prior to the experiments, the mice were adapted to the dark for 1.5 h and then anesthetized with 1% pentobarbital (0.01 ml/g body weight) injected intraperitoneally. The mice were immobilized in a custom-made animal holder to minimize movements caused by respiration and heartbeat. The pupil was fully dilated with 2% tropicamide and 2% phenylephrine hydrochloride, and the cornea was moisturized with 0.1% sodium hyaluronate solution to eliminate the influence of tear film evaporation. The animal’s body temperature was maintained using a heating pad throughout the imaging session [21]. All fOCTA scans were conducted before noon to suppress the influence of diurnal variation. All animal experimental procedures were approved by the Animal Care and Use Committee of Zhejiang University (ZJU20220134).
2.2 System setup
The fOCTA system is mainly composed of two modules (Fig. S1): retinal functional stimulation (i.e., FLS) and functional hyperemia monitoring (i.e., time-lapse OCTA). Diffused FLS was generated by a green light-emitting diode with a central wavelength of 520 nm. In the dark room, the illuminance of FLS was 1000 lux with a 50 ms pulse width and a 10 Hz repetition rate of pulse trains. Each FLS trial consisted of a 30 s baseline period, followed by 30 s FLS and a subsequent 100 s post-FLS period that allowed full recovery to baseline.
FLS-evoked retinal functional hyperemia was recorded with time-lapse OCTA, which is detailed in previous study [22]. In brief, the custom-built prototype system was based on a spectral domain configuration. A superluminescent diode with a central wavelength of 840 nm and a spectral bandwidth of 100 nm was used as the light source, yielding a measured axial resolution of ∼4 µm in air. The probe beam was focused on the retina with a lateral resolution of ∼10 µm. The spectral interferograms were recorded by a line-scan camera (e2v, UK) at a scan rate of 120 kHz. The time-lapse OCTA imaging was synchronized to the FLS via a triggering circuit. The total light power of OCT and FLS on the pupil is ∼1 mW, which is within the American National Standards Institute safety limit [23].
2.3 Data acquisition
Dynamic volumetric OCTA was conducted with a repeated stepwise raster scanning protocol (z-x-y). Each OCTA volume was acquired within 2 s with 256 A-lines (x) per B-scan, 256 tomographic positions (y) per volume and 3 repeated B-scans at each position. Repeated volumetric OCTA scans were centered on the optical nerve head with an FOV of 2 mm × 2 mm (x-y) and performed at a time interval of 6 s (20 s in the post-FLS period) to record the time course of the FLS-evoked hyperemic response. Notably, only the baseline and FLS periods (for a total of 60 s) were recorded by time-lapse OCTA to further shorten the examination time of DM mice.
2.4 Data processing
In each OCTA volume, the raw spectral interferogram was preprocessed with k-linearization and dispersion compensation and then Fourier transformed to generate the OCT structure [24], and the inverse signal-to-noise ratio and decorrelation OCT angiography (ID-OCTA) algorithm was used to create the OCT angiogram (Fig. S2) [17]. The retinal layers were segmented with a graph search algorithm [25]. Accordingly, en face angiograms were generated by projecting the OCTA signal within specific retinal slabs, including the superficial vascular plexus (SVP), intermediate capillary plexus (ICP) and deep capillary plexus (DCP). To improve the accuracy of functional hyperemia quantification, the artifacts of retinal major vessels on ICP and DCP were subtracted based on its intensity-normalized decorrelation values [26]. The OCTA data processing was accelerated with a graphics processing unit (GPU) for a real-time en face angiogram display; therefore, the instant feedback of angiogram quality ensured reliable stabilization of the animals and a lack of apparent spatial motion in the OCTA datasets [22].
2.5 Functional hyperemia quantification
Retinal functional hyperemia was quantified with the percentage changes in vessel caliber (ΔVC), vessel density (ΔVD), flow decorrelation average (ΔFDa) and flow decorrelation sum (ΔFDs) during FLS. Each angiogram slab of the same trial was binarized with a consistent threshold, i.e., the Otsu threshold of the averaged baseline angiogram [22]. The width of major vessels (i.e., arterioles and venules) was defined as the vessel caliber (VC), and the percentage area of capillaries was defined as the vessel density (VD). Each complex decorrelation matrix was masked with its binarized angiogram to totally clear the background [17], and the average and sum of the decorrelation values in specific vessels were defined as the flow decorrelation average (FDa) and flow decorrelation sum (FDs), respectively. Then, the percentage changes ΔVC, ΔVD, ΔFDa and ΔFDs of each vessel during the FLS trial were calculated. The average responses of same kind of vessels were applied to generate the corresponding time courses. The mean value during the FLS period (average over 18, 24 and 30 s after FLS onset) was used as the index of functional hyperemia. In addition, the responses of arterioles and venules were further classified into dilation (mean ΔVC > 0) and constriction (mean ΔVC < 0). The FDs is suggested to be an index of total blood flow, while the FDa indicates average blood flux because decorrelation estimation is related to flow velocity and hematocrit [27,28]. Data were analyzed by two-tailed Student t test to compare variables between two groups, and P $< $ 0.05 was considered statistically significant.
2.6 Functional hyperemia imaging
First, the en face angiograms were averaged separately over the baseline (all 5 timepoints) and FLS (last 3 timepoints) periods to improve the signal-to-noise ratio. Second, the average angiograms were binarized and skeletonized [29] to collapse the vascular caliber and preserve the connectivity information. Then, each vascular segment was located by removing the bifurcation points (number of neighboring vascular pixels $\ge $ 3) from the skeleton. Finally, the percentage change ΔFDs of each capillary segment was calculated and used as the contrast of the fOCTA images.
2.7 Immunolabeling
Both the DM mice without AG treatment and age-matched controls were lethally injected with pentobarbital for euthanasia. For retinal whole-mounts, mouse eyeballs were fixed in 4% paraformaldehyde for 1 h at room temperature, then the whole retinas were dissected. Next, the retinas were blocked in 5% donkey serum for 1 h and permeabilized with 0.5% Triton for 30 min. The retinas were then incubated with IB4 (Sigma) and primary antibodies against NG2 (Abcam) and glial fibrillary acid protein (GFAP, Proteintech) overnight at 4 °C to immunolabel endothelial cells, pericytes and GFAP, respectively. Secondary antibodies (Thermo Fisher) were applied for 1 h at room temperature. The slides were imaged and the density of immunolabeled cells or protein was quantified.
For retinal sections, sagittal planes sections close to the optic nerve head were cut at a thickness of 5 µm. After deparaffinization, the antigen retrieval was performed with heated EDTA. The sections were then incubated with primary antibody against inducible nitric oxide synthase (iNOS) (Proteintech) overnight at 4 °C. Secondary antibodies (Thermo Fisher) were applied for 1 h at room temperature. The stained sections were imaged and the density of iNOS were quantified. Apoptotic neurons were detected by TUNEL. Briefly, the deparaffinized sections were permeabilized with PBS (pH 7.4) containing Triton, then incubated in a reaction mixture solution for 60 min at 37 °C. The cell nucleus was stained by DAPI. Finally, TUNEL positive cells per eye were manually quantified.
3. Results
3.1 Functional hyperemia response to FLS in the control retina
With the proposed fOCTA, the time courses of retinal functional hyperemia were recorded in synchronization with FLS, and apparent hyperemic responses (vessel vasodilation and blood flow increase) were observed in all three retinal vascular plexuses in normal mice (see the en face angiographic video in Visualization 1 and Fig. 1). The major vessels (i.e., arterioles and venules) of the retina dilated during FLS (Figs. 1(A)–1(B), insets I-II and i) and then gradually returned to baseline levels after the cessation of stimulus (Fig. 1(C), inset III and i). Significant increases in VC and FDs (P < 0.05 between FLS on and off at 30 s) were observed in response to FLS on major vessels (Fig. 1(J)). Similarly, significant increases in VD and FDs (P < 0.05 between FLS on and off at 30 s) were observed in the capillaries of all three plexus layers: the capillaries of SVP (Figs. 1(A)–1(C), insets IV-VI and ii, and Fig. 1(K), ICP (Figs. 1(D)–1(F), insets VII-IX and iii, and Fig. 1(L) and DCP (Figs. 1(G)–1(I), insets X-XII and iv, and Fig. 1(M).
3.2 Functional hyperemia of retinal arterioles and venules in DM mice before and after AG treatment
Both arterioles and venules presented apparent functional hyperemia loss in DM mouse retinas, and obvious restoration after AG treatment. Compared with that of the age-matched controls, the FLS-evoked arteriole dilation ΔVC was decreased to 2.0% (5.4% in control, P < 0.05) in DM mice, and then restored to 2.6% (no significance) after AG treatment (Figs. 2(A) and 2(C)). The arteriole constriction ΔVC was also observed during FLS, and it was decreased to -3.1% (-1.5% in control, no significance) in DM mice, and then increased to -2.3% (no significance) after AG treatment (Figs. 2(B) and 2(C)). Notably, significant changes were found in the incidence of arteriole constriction which was increased to 63.9% (13.0% in control, P < 0.001) in DM retinas, and then decreased to 25.0% (P < 0.05) after AG treatment (Fig. 2(D)). Similar hyperemic responses were observed in the retinal venules. The venule dilation ΔVC was decreased to 2.4% (5.5% in control, no significance) in DM mice, and then changed to 2.3% (no significance) after AG treatment (Figs. 2(E) and 2(G). The venule constriction ΔVC was decreased to -2.7% (-1.2% in control, no significance) in DM mice, and then increased to -1.3% (no significance) after AG treatment (Figs. 2(F) and 2(G). The incidence of FLS-evoked venule constriction was increased to 51.3% (12.0% in control, P < 0.05) in DM mice, and then decreased to 21.0% (no significance) after AG treatment (Fig. 2(H).
Although both arterioles (Figs. 2(I)–2(J)) and venules (Figs. 2(K)–2(L) presented a clear attenuation of the FLS-evoked blood flow increase (ΔFDs) in DM mouse retinas and a certain restoration after AG treatment, no significance was observed. Similar behaviors were also observed in the changes of ΔFDa (Figs. S3A-S3F). Notably, the decorrelation metrics (FDa and FDs) may not be valid for the blood flow of major vessels because of the saturation limit: their FDa was measured at ∼0.42 at baseline (Figs. S4C and S4F) and ∼0.43 in FLS (Figs. S3A-S3F), both of which approach the decorrelation saturation limit of 0.56 ± 0.20 for a kernel size of 5 [30].
3.3 Functional hyperemia of retinal capillaries in DM mice before and after AG treatment
The SVP capillaries exhibited significant functional hyperemia loss in DM mouse retinas, and significant restoration after AG treatment. Specifically, the FLS-evoked vasodilation ΔVD was attenuated to -3.0% (4.9% in control, P < 0.001) in DM mice, and then restored to 2.6% (P < 0.05) after AG treatment (Figs. 3(A)–3(B)). The FLS-evoked blood flow ΔFDs was decreased to -1.7% (9.4% in control, P < 0.001) in DM group, and then restored to 3.4% (P < 0.01) after AG treatment (Figs. 3(C)–3(D)). In addition, the blood flow index ΔFDa of SVP capillaries presented a downward trend in DM and a recovery after AG treatment, but no significance was observed (Figs. S3G-S3H).
In particular, the ICP exhibited the most dramatic functional hyperemia loss in DM mouse retinas and significant restoration after AG treatment. Specifically, the FLS-evoked vasodilation ΔVD was attenuated to -3.4% (9.5% in control, P < 0.001) in DM group, and then restored to 2.7% (P < 0.05) after AG treatment (Figs. 3(E)–3(F)). The FLS-evoked blood flow ΔFDs was decreased to -2.9% (13.0% in control, P < 0.001) in DM group, and then restored to 3.8% (P < 0.05) after AG treatment (Figs. 3(G)–3(H). The ΔFDa of ICP presented a downward trend in DM and a weak recovery after AG treatment, but no significance was observed (Figs. S3I-S3J).
In addition, the DCP presented significant functional hyperemia loss in DM mouse retinas and apparent restoration after AG treatment. Specifically, the FLS-evoked vasodilation ΔVD was decreased to -3.7% (2.3% in control, P < 0.05) in DM group, and then restored to 1.8% (P < 0.05) after AG treatment (Figs. 3(I)–3(J)). The FLS-evoked blood flow ΔFDs was attenuated to -2.5% (5.6% in control, P < 0.05) in DM group, and then increased to 2.0% (no significance) after AG treatment (Figs. 3(K)–3(L). The ΔFDa of DCP presented a significant attenuation (P < 0.05) in DM and a weak recovery after treatment (Figs. S3K-S3L).
3.4 Retinal functional hyperemia imaging
A retinal fOCTA image (Fig. 4), i.e., a depth-resolved map of the functional hyperemia in the retina, was generated with the contrast of the functional index ΔFDs, which had the largest changes in response to FLS. Distinct from conventional angiographic images (Figs. 4(A)–4(I)), the retinal functional images showed obvious differences between the normal and DM mice. Generally, most vessels exhibited apparent functional hyperemia in response to FLS in normal mice (Figs. 4(J)–4(L), significant loss of functional hyperemia in DM mice (Figs. 4(M)–4(O)) and obvious restoration of functional hyperemia after AG treatment (Figs. 4(P)–4(R)).
3.5 Pathological changes in the DM retina
Particularly, the iNOS was upregulated in DM retina than control groups (DM: 68.3%, control: 19.5%, P < 0.001) (Figs. 5(A)–5(C)). In addition, a significant increase in GFAP expression was also observed in DM retina (DM: 33.0%, control: 7.1%, P < 0.001) (Figs. 5(D)–5(F)). To compare the microvasculopathy in the DM and age-matched control groups, the density of immunolabeled vascular cells was analyzed. Hoverer, there was no significant loss of endothelial cells (DM: 14.4%; control: 13.4%) (Figs. 5(G)–5(I)) or pericytes (DM: 4.4%; control: 4.7%) (Figs. 5(J)–5(L) in DM retina. The retinal neuronal death was directly detected by TUNEL staining. There was no difference in the number of TUNEL positive cells between DM (0.88 cells per eye) and age-matched control (0.75 cells per eye) groups (Figs. 5(M)–5(O)). Besides, no significant retinal thinning was observed in the thickness from the nerve fiber layer to the ganglion cell layer (DM: 26.6 µm; control: 26.1 µm) and the thickness of inner nuclear layer (DM: 25.9 µm; control: 26.2 µm) (Figs. 5(P)–5(S)). The measured thickness was a mean of the inner ring with diameters of 0.6 to 1.8 mm centered on the optic nerve head [31].
4. Discussion
Assessing capillary functional hyperemia with OCTA is desirable but still challenging [11,20]. First, capillaries have a small caliber and might exhibit a great deal of spatial misalignment due to bulk motion during dynamic monitoring, posing a great challenge to directly comparing the capillary changes before and after functional stimulation. The existing OCTA registration methods were developed with the aim of entirely eliminating the changes between the OCTA images acquired from the same space [32,33]; thus, they may not be applicable to the current fOCTA imaging, which requires preserving the dynamic changes in functional hyperemia. Second, the transient functional hyperemia response is highly dependent on the dose of FLS [34] and the recovery time if measured after FLS cessation (Figs. 1(J)–1(M), which should be precisely controlled in quantifying the percentage changes in vessel and blood flow. Most likely due to the fluctuations caused by the spatial misalignment and the recovery time, no significance was observed in the retinal hyperemic response to FLS in previous report [11], which compared the en face angiograms acquired before and after a fixed 20 s dose of FLS with a commercial OCTA system. In our fOCTA, a high-quality retinal angiogram time series can be acquired in well-stabilized mice without apparent spatial bulk motion (see Visualization 1), owing to the instant feedback of angiogram quality enabled by the real-time data processing in the GPU-accelerated ID-OCTA. In addition, rigid synchronization between the OCTA imaging and FLS was employed in fOCTA to suppress the irradiation dose-induced fluctuations in the functional measurements. Accordingly, 3D retinal functional hyperemia imaging was achieved with capillary resolution by effectively extracting the FLS-evoked hyperemic response from each capillary segment.
Our fOCTA has strong potential for detecting the early-stage DR. The observed functional arteriole and venule hyperemia in normal mouse retina and the loss of the functional hyperemia in early-stage DR with our fOCTA were generally compatible with the reported observations in humans and animal models with different methods in the literatures. As measured in human subjects with DVA, a flicking light typically evokes a 3.6-4.5% dilation in arteries and a 2.0-4.5% dilation in veins [8,35], which are reduced by ∼60% in arteries and ∼30% in veins of patients with both type 1 and type 2 diabetes [8,36,37]. In rat studies using confocal microscopy, flicking light evokes a 10.8% dilation in arterioles of control animals, and these dilations are reduced by 61% in the streptozotocin rat model of type 1 diabetes [38]. Our fOCTA measured ∼5.4 and 5.5% dilations in arterioles and venules of control mice, respectively, and ∼63.0 and 56.4% reductions in arterioles (Fig. 2(C)) and venules (Fig. 2(G) of streptozotocin-induced diabetic mice, respectively. The discrepancies in our measurements may be attributable to the differences in species, stimulation protocols and imaging tools. In addition to the reduction of the vasodilation amplitude, the incidence of FLS-evoked vasoconstriction also increased to ∼63.9% (13.0% in control) in retinal arterioles of DM mice in vivo (Fig. 2(D)), which was in accordance with the observed 53% incidence of arteriolar constriction in isolated DM retinas [39]. In their study [39], iNOS was upregulated in retinal neurons and glia in diabetic retinas, and retinal nitric oxide levels were increased [40]. High nitric oxide levels reduce vasodilation and enhance vasoconstriction in arterioles [41]. Thus, the increased nitric oxide levels resulting from iNOS upregulation (Figs. 5(A)–5(C)) might be responsible for the loss of functional hyperemia observed in DM retinal arterioles with fOCTA. In brief, our fOCTA allows a precise and effective measurement of functional hyperemia in all arterioles and venules with a single trial of FLS.
In our fOCTA, the retinal capillaries, particularly the ICP, presented apparent functional hyperemia in normal mouse and significant functional hyperemia loss in the early-stage DR. Our fOCTA found a more pronounced functional hyperemia response to FLS in the ICP than in the SVP capillaries and DCP (ΔVD: ∼9.5% vs. 4.9% (P < 0.05) and 2.3% (P < 0.01), respectively; ΔFDs: ∼13.0% vs. 9.4% (no significance) and 5.6% (P < 0.05), respectively), which was in agreement with previous findings [10,14]. In the retina, Ca2+ increases in Müller cells are both necessary and sufficient for capillary dilation, and the ICP, the only vascular layer where FLS generates Ca2+ responses in Müller cell endfeet, is the only layer where capillaries dilate in mouse eyecups [14]. In addition, FLS activates the most dramatic changes in ICP to supply the highest metabolic demand of neuronal somata and synapses in the inner retina [10], which is distinct from the dark adaptation that evokes the largest responses in DCP [11,42]. In contrast, the most pronounced FLS-evoked functional hyperemia was found in SVP capillaries in humans using a commercial OCTA system and an independent FLS module [20]. The discrepancy may be due to the species differences as well as the unknown irradiation dose and registration method used in their study [20].
In our fOCTA, the ICP also exhibited a larger magnitude of functional hyperemia loss than SVP capillaries or DCP in terms of both the vasodilation ΔVD (a decrease of ∼12.9 vs. 7.9 and 6.0%, respectively) and blood flow increase ΔFDs (a decrease of ∼15.9 vs. 11.1 and 8.1%, respectively) from the control to DM groups, as well as a high reliability in detecting the loss of functional hyperemia in DM: vasodilation and blood flow increase exhibited the most significant abnormality (P < 0.001) as early as 4 wk of DM duration. This is mostly because ICP functional hyperemia is mediated by Müller cells [14], and Müller cells undergo an upregulation of GFAP (Figs. 5(D)–5(F)), indicating the beginning stages of reactive gliosis [43]. In addition, prior studies have also shown that the ICP is more likely affected in DR with more diabetic vasculopathy, such as microaneurysms and capillary loops [44,45]. Thus, our findings indicate that the diminished ICP functional hyperemia response may reflect early retinal NVC dysfunction and have the potential to be a sensitive biomarker for detecting early DR.
In this work, the loss of functional hyperemia was observed at an early stage of DR with few overt signs of retinopathy. There was no significant loss of retinal neurons, endothelial cells and pericytes (Figs. 5(G)–5(S)), and no significant alteration in the baselines of retinal vascular morphology and blood flow (Fig. S4) at 4 wk of DM duration. The loss of capillary functional hyperemia was earlier than the reported neurodegeneration at 6 wk by measuring the combined thickness of the nerve fiber layer and ganglion cell layer [46] and much earlier than the reported vasculopathy at 24 wk by counting the pericyte loss and acellular capillaries [47]. The loss of functional hyperemia will create a mismatch between energy supply and demand, depriving neurons of oxygen and nutrients [48]. Thus, functional hyperemia loss may be an important contributing factor in the onset and development of DR [37]. Despite recent advances, the mechanism mediating the functional hyperemia response and its loss in DM remain unclear. Retinal fOCTA, capable of mapping functional hyperemia at the single-capillary level, would provide new insights into the pathophysiology, screening and treatment of early DR.
In addition to measuring the loss of functional hyperemia in early DR, fOCTA can also evaluating the recovery of NVC function during the DR treatment. The upregulated iNOS (Figs. 5(A)–5(C)) may inhibit the glial release of dilatory agents and disrupt the NVC process in DM retina [38,39]. Acute AG injection can inhibit the iNOS and reduce the nitric oxide level in retina, thereby restoring the normal signal transduction between neuronal activity and vasodilation [38]. With the fOCTA, we clearly observed that the arteriolar dilation ΔVC was restored and the incidence of FLS-evoked arteriolar constriction was reduced to the control level after the acute treatment with iNOS inhibitor, which were in accordance with the reported restoration of FLS-evoked arteriole response in DM retina in vitro [39]. In addition to the major vessels, the treatment-induced recovery of the normal functional hyperemia was observed with fOCTA in all the three capillary plexuses. Thus, fOCTA has strong potential for guiding DR treatment.
In our fOCTA, the single-capillary resolution allows a clear visualization of the regional diversity of capillary hyperemia responses to FLS, e.g., the negative (or positive) responses in normal (or DM) mice in Fig. 4. The regional diversity in the normal retina might be attributed to the fact that only a portion of the retinal neurons can be concurrently activated, leading to a nonuniform energy requirement of neurons within the entire neural tissue [49]. In addition, the vasoconstriction in nonactivated neural regions may be a part of the functional hyperemia responses and may play a role in the redistribution of retinal blood flow to meet the energy demand of the activated neural regions [50,51]. In particular, the regional diversity of the functional hyperemia loss and restoration in the DM retinas might correlate to the heterogeneous microvasculopathy in the advanced DR [52], which deserves a further investigation in the future.
Our fOCTA has several advantages over the existing FLS-based methods for assessing retinal NVC function. First, the conventional DVA can assess functional hyperemia only in major vessels and requires averaging over multiple FLS trials (typically requiring a total time of ∼350 s, including 50 s baseline, followed by 3 trials consisting of 20 s FLS and 80 s recovery) [13]. Benefiting from the capillary resolution of OCTA, our fOCTA extended the measurement of functional hyperemia to capillaries under a single-trial FLS (30 s baseline and 30 s FLS). Rather than the global function revealed by arteriole and venule functional hyperemia, the capillary hyperemia provides information on FLS response of specific vascular plexus and regional NVC alterations in DR. The single-trial FLS could eliminate the recovery period between the repeated trials and dramatically shorten the imaging time to tens of seconds, which would greatly improve the subject tolerance. Second, in contrast to confocal imaging [10], fOCTA increased the FOV from hundreds of micrometers to a millimeter scale. fOCTA utilizes the intrinsic motion contrast from flowing red blood cells, avoiding the need for exogenous contrast, and uses infrared light in the imaging channel, avoiding disruptive stimulation. As a noninvasive examination, fOCTA would be convenient for clinical application and longitudinal study.
Notably, the current GPU-accelerated fOCTA method could effectively avoid the motion-induced spatial mismatch during the imaging session of anesthetized and stabilized animals, but the spatial mismatch induced by ocular bulk motion in awake humans poses a great challenge to the clinical practice of fOCTA. In addition, the eye tracking method used in the commercial OCTA system may not be applicable to the fOCTA system because tracking would cause an apparent variance in acquisition time (i.e., irradiation dose) of each volumetric OCTA data point [53] during FLS, resulting in a large fluctuation in functional hyperemia measurement. Thus, a sophisticated image registration method is needed for fOCTA imaging of the human eye, not only achieving spatial registration of blood vessels but also preserving information on vascular dynamic changes.
In summary, we acquired high-resolution, time-lapse 3D OCTA images of mouse retinas in synchronization with FLS and calculated the functional hyperemia response along capillary segments with metrics of vasodilation (ΔVC or ΔVD) and blood flow increase (ΔFDa and ΔFDs). Using the developed fOCTA, we achieved label-free 3D imaging of functional hyperemia across the entire retinal vascular tree at the single-capillary level in the mouse retina. The retinal capillaries, in particular the ICP, exhibited apparent hyperemic response to FLS in normal mice, and significant functional hyperemia loss (P < 0.001) at an early stage of DR with few overt signs of retinopathy and restoration after AG treatment. Use of our fOCTA method to generate 3D maps of functional hyperemia at the single-capillary level will help to clarify and expand our understanding of retinal NVC function in health and disease, particularly providing new insights into the pathophysiology, screening and treatment of early DR.
Funding
National Natural Science Foundation of China (62075189, T2293751, 82201197, 62035011, 11974310, 31927801); Zhejiang Provincial Natural Science Foundation of China (LR19F050002); National Key Research and Development Program of China (2017YFA0700501); MOE Frontier Science Center for Brain Science & Brain-Machine Integration, Zhejiang University.
Disclosures
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. D. S. Fong, L. P. Aiello, F. L. Ferris, and R. Klein, “Diabetic retinopathy,” Diabetes Care 27(10), 2540–2553 (2004). [CrossRef]
2. S. Wild, G. Roglic, A. Green, R. Sicree, and H. King, “Global prevalence of diabetes - Estimates for the year 2000 and projections for 2030,” Diabetes Care 27(5), 1047–1053 (2004). [CrossRef]
3. N. Cheung, P. Mitchell, and T. Y. Wong, “Diabetic retinopathy,” Lancet 376(9735), 124–136 (2010). [CrossRef]
4. V. Muoio, P. B. Persson, and M. M. Sendeski, “The neurovascular unit - concept review,” Acta Physiol. 210(4), 790–798 (2014). [CrossRef]
5. D. A. Antonetti, R. Klein, and T. W. Gardner, “Mechanisms of disease diabetic retinopathy,” N. Engl. J. Med. 366(13), 1227–1239 (2012). [CrossRef]
6. G. Gerhard, J. Chua, B. Y. Tan, D. Wong, D. Schmidl, and L. Schmetterer, “Retinal neurovascular coupling in diabetes,” J. Clin. Med. 9(9), 2829 (2020). [CrossRef]
7. T. Y. Wong, C. M. G. Cheung, M. Larsen, S. Sharma, and R. Simo, “Diabetic retinopathy,” Nat. Rev. Dis. Primers2, (2016).
8. J. Mandecka, M. Dawczynski, N. Blum, C. Muller, G. Kloos, W. Wolf, H. Vilser, U. A. Hoyer, and Muller, “Influence of flickering light on the retinal vessels in diabetic patients,” Diabetes Care 30(12), 3048–3052 (2007). [CrossRef]
9. T. T. Nguyen, J. Shaw, R. Kawasaki, W. Vilser, J. J. Wang, T. Y. Wong, and A. J. Kreis, “Flicker light-induced retinal vasodilation in diabetes and diabetic retinopathy,” Diabetes Care 32(11), 2075–2080 (2009). [CrossRef]
10. T. E. Kornfield and E. A. Newman, “Regulation of blood flow in the retinal trilaminar vascular network,” J. Neurosci. 34(34), 11504–11513 (2014). [CrossRef]
11. P. L. Nesper, H. E. Lee, A. E. Fayed, G. W. Schwartz, F. Yu, and A. A. Fawzi, “Hemodynamic response of the three macular capillary plexuses in dark adaptation and flicker stimulation using optical coherence tomography angiography,” Invest. Ophthalmol. Visual Sci. 60(2), 694–703 (2019). [CrossRef]
12. Y. Cheung, M. K. Ikram, R. Klein, and T. Y. Wong, “The clinical implications of recent studies on the structure and function of the retinal microvasculature in diabetes,” Diabetologia 58(5), 871–885 (2015). [CrossRef]
13. M. Lim, M. B. Sasongko, M. K. Ikram, E. Lamoureux, J. J. Wang, T. Y. Wong, and C. Y. Cheung, “Systemic associations of dynamic retinal vessel analysis: a review of current literature,” Microcirculation 20(3), 257–268 (2013). [CrossRef]
14. K. R. Biesecker, A. I. Srienc, A. M. Shimoda, A. Agarwal, D. E. Bergles, P. Kofuji, and E. A. Newman, “Glial Cell Calcium Signaling Mediates Capillary Regulation of Blood Flow in the Retina,” J. Neurosci. 36(36), 9435–9445 (2016). [CrossRef]
15. R. K. Wang, S. L. Jacques, Z. Ma, S. Hurst, S. R. Hanson, and A. Gruber, “Three dimensional optical angiography,” Opt. Express 15(7), 4083–4097 (2007). [CrossRef]
16. Y. L. Jia, O. Tan, J. Tokayer, B. Potsaid, Y. M. Wang, J. J. Liu, M. F. Kraus, H. Subhash, J. G. Fujimoto, J. Hornegger, and D. Huang, “Split-spectrum amplitude-decorrelation angiography with optical coherence tomography,” Opt. Express 20(4), 4710–4725 (2012). [CrossRef]
17. L. Z. Huang, Y. M. Fu, R. X. Chen, S. S. Yang, H. X. Qiu, X. N. Wu, S. Y. Zhao, Y. Gu, and P. Li, “SNR-Adaptive OCT angiography enabled by statistical characterization of intensity and decorrelation with multi-variate time series model,” IEEE Trans. Med. Imaging 38(11), 2695–2704 (2019). [CrossRef]
18. H. Kashani, C. L. Chen, J. K. Gahm, F. Zheng, G. M. Richter, P. J. Rosenfeld, Y. G. Shi, and R. K. K. Wang, “Optical coherence tomography angiography: A comprehensive review of current methods and clinical applications,” Prog. Retinal Eye Res. 60, 66–100 (2017). [CrossRef]
19. T. Son, M. Alam, D. Toslak, B. Q. Wang, Y. M. Lu, and X. C. Yao, “Functional optical coherence tomography of neurovascular coupling interactions in the retina,” J. Biophotonics 11(12), e201800089 (2018). [CrossRef]
20. M. Kallab, N. Hommer, B. Y. Tan, M. Pfister, A. Schlatter, R. M. Werkmeister, J. Chua, D. Schmidl, L. Schmetterer, and G. Garhofer, “Plexus-specific effect of flicker-light stimulation on the retinal microvasculature assessed with optical coherence tomography angiography,” Am. J. Physiol-Heart C. 320(1), H23–H28 (2021). [CrossRef]
21. K. Y. Liu, T. P. Zhu, L. Yao, Z. Y. Zhang, H. K. Li, J. Ye, and P. Li, “Noninvasive OCT angiography-based blood attenuation measurements correlate with blood glucose level in the mouse retina,” Biomed. Opt. Express 12(8), 4680–4688 (2021). [CrossRef]
22. X. Deng, K. Liu, T. Zhu, D. Guo, X. Yin, L. Yao, Z. Ding, J. Ye, and P. Li, “Dynamic inverse SNR-decorrelation OCT angiography with GPU acceleration,” Biomed. Opt. Express 13(6), 3615–3628 (2022). [CrossRef]
23. ANSI, “American National Standard for Safe Use of Lasers,” ANSI Z136.1–2007 (Laser Institute of America, Orlando, 2007) pp. 1–249.
24. M. Wojtkowski, V. J. Srinivasan, T. H. Ko, J. G. Fujimoto, A. Kowalczyk, and J. S. Duker, “Ultrahigh-resolution, high-speed, Fourier domain optical coherence tomography and methods for dispersion compensation,” Opt. Express 12(11), 2404–2422 (2004). [CrossRef]
25. S. J. Chiu, X. T. Li, P. Nicholas, C. A. Toth, J. A. Izatt, and S. Farsiu, “Automatic segmentation of seven retinal layers in SDOCT images congruent with expert manual segmentation,” Opt. Express 18(18), 19413–19428 (2010). [CrossRef]
26. M. Zhang, T. S. Hwang, J. P. Campbell, S. T. Bailey, D. J. Wilson, D. Huang, and Y. L. Jia, “Projection-resolved optical coherence tomographic angiography,” Biomed. Opt. Express 7(3), 816–828 (2016). [CrossRef]
27. Y. L. Jia, S. T. Bailey, T. S. Hwang, S. M. McClintic, S. S. Gao, M. E. Pennesi, C. J. Flaxel, A. K. Lauer, D. J. Wilson, J. Hornegger, J. G. Fujimoto, and D. Huang, “Quantitative optical coherence tomography angiography of vascular abnormalities in the living human eye,” Proc. Natl. Acad. Sci. U. S. A. 112(18), E2395–E2402 (2015). [CrossRef]
28. D. Pechauer, T. S. Hwang, A. M. Hagag, L. Liu, O. Tan, X. B. Zhang, M. Parker, D. Huang, D. J. Wilson, and Y. L. Jia, “Assessing total retinal blood flow in diabetic retinopathy using multiplane en face Doppler optical coherence tomography,” Br. J. Ophthalmol. 102(1), 126–130 (2018). [CrossRef]
29. Y. M. Zhang, H. K. Li, T. T. Cao, R. X. Chen, H. X. Qiu, Y. Gu, and P. Li, “Automatic 3D adaptive vessel segmentation based on linear relationship between intensity and complex-decorrelation in optical coherence tomography angiography,” Quant. Imag. Med. Surg. 11(3), 895–906 (2021). [CrossRef]
30. R. X. Chen, L. Yao, K. Y. Liu, T. T. Cao, H. K. Li, and P. Li, “Improvement of decorrelation-based OCT angiography by an adaptive spatial-temporal kernel in monitoring stimulus-evoked hemodynamic responses,” IEEE Trans. Med. Imaging 39(12), 4286–4296 (2020). [CrossRef]
31. ETDRS Research Group, “Early photocoagulation for diabetic retinopathy. Early Treatment Diabetic Retinopathy Study report number 9,” Ophthalmology 98(5 Suppl), 766–785 (1991).
32. W. Wei, A. J. Deegan, and R. K. K. Wang, “Automatic motion correction for in vivo human skin optical coherence tomography angiography through combined rigid and nonrigid registration,” J. Biomed. Opt. 22(6), 1 (2017). [CrossRef]
33. J. Wang, A. Camino, X. H. Hua, L. Liu, D. Huang, T. S. Hwang, and Y. L. Jia, “Invariant features-based automated registration and montage for wide-field OCT angiography,” Biomed. Opt. Express 10(1), 120–136 (2019). [CrossRef]
34. B. Tan, E. Mason, B. MacLellan, and K. K. Bizheva, “Correlation of visually evoked functional and blood flow changes in the rat retina measured with a combined OCT + ERG System,” Invest. Ophthalmol. Visual Sci. 58(3), 1673–1681 (2017). [CrossRef]
35. K. Polak, L. Schmetterer, and C. E. Riva, “Influence of flicker frequency on flicker-induced changes of retinal vessel diameter,” Invest. Ophthalmol. Visual Sci. 43(8), 2721–2726 (2002).
36. B. Pemp, G. Garhofer, G. Weigert, K. Karl, H. Resch, M. Wolzt, and L. Schmetterer, “Reduced retinal vessel response to flicker stimulation but not to exogenous nitric oxide in type 1 diabetes,” Invest. Ophthalmol. Visual Sci. 50(9), 4029–4032 (2009). [CrossRef]
37. R. Nippert and E. A. Newman, “Regulation of blood flow in diabetic retinopathy,” Vis. Neurosci. 37, E004 (2020). [CrossRef]
38. E. A. Mishra and Newman, “Aminoguanidine reverses the loss of functional hyperemia in a rat model of diabetic retinopathy,” Front. Neuroenerg. 3, 10 (2011). [CrossRef]
39. Mishra and E. A. Newman, “Inhibition of inducible nitric oxide synthase reverses the loss of functional hyperemia in diabetic retinopathy,” Glia 58(16), 1996–2004 (2010). [CrossRef]
40. Y. P. Du, M. A. Smith, C. M. Miller, and T. S. Kern, “Diabetes-induced nitrative stress in the retina, and correction by aminoguanidine,” J. Neurochem. 80(5), 771–779 (2002). [CrossRef]
41. M. R. Metea and E. A. Newman, “Glial cells dilate and constrict blood vessels: A mechanism of neurovascular coupling,” J. Neurosci. 26(11), 2862–2870 (2006). [CrossRef]
42. Y. S. Zhang, I. Mucollari, C. C. Kwan, G. Dingillo, J. Amar, G. W. Schwartz, and A. A. Fawzi, “Reversed neurovascular coupling on optical coherence tomography angiography is the earliest detectable abnormality before clinical diabetic retinopathy,” J. Clin. Med. 9(11), 3523 (2020). [CrossRef]
43. V. K. Lee, B. M. Hosking, J. Holeniewska, E. C. Kubala, P. L. von Leithner, P. J. Gardner, R. H. Foxton, and D. T. Shima, “BTBR ob/ob mouse model of type 2 diabetes exhibits early loss of retinal function and retinal inflammation followed by late vascular changes,” Diabetologia 61(11), 2422–2432 (2018). [CrossRef]
44. W. Choi, N. K. Waheed, E. M. Moult, M. Adhi, B. Lee, T. De Carlo, V. Jayaraman, C. R. Baumal, J. S. Duker, and J. G. Fujimoto, “Ultrahigh Speed Swept Source Optical Coherence Tomography Angiography of Retinal and Choriocapillaris Alterations in Diabetic Patients with and without Retinopathy,” Retina 37(1), 11–21 (2017). [CrossRef]
45. J. Moore, S. Bagley, G. Ireland, D. McLeod, and M. E. Boulton, “Three dimensional analysis of microaneurysms in the human diabetic retina,” J. Anat. 194(1), 89–100 (1999). [CrossRef]
46. H. Sohn, H. W. van Dijk, C. H. Jiao, P. H. B. Kok, W. Jeong, N. Demirkaya, A. Garmager, F. Wit, M. Kucukevcilioglu, M. E. J. van Velthoven, J. H. DeVries, R. F. Mullins, M. H. Kuehn, R. O. Schlingemann, M. Sonka, F. D. Verbraak, and M. D. Abramoffa, “Retinal neurodegeneration may precede microvascular changes characteristic of diabetic retinopathy in diabetes mellitus,” Proc. Natl. Acad. Sci. U. S. A. 113(19), E2655–E2664 (2016). [CrossRef]
47. R. A. Feit-Leichman, R. Kinouchi, M. Takeda, Z. G. Fan, S. Mohr, T. S. Kern, and D. F. Chen, “Vascular damage in a mouse model of diabetic retinopathy: Relation to neuronal and glial changes,” Invest. Ophthalmol. Visual Sci. 46(11), 4281–4287 (2005). [CrossRef]
48. R. F. Gariano and T. W. Gardner, “Retinal angiogenesis in development and disease,” Nature 438(7070), 960–966 (2005). [CrossRef]
49. P. Lennie, “The cost of cortical computation,” Curr. Biol. 13(6), 493–497 (2003). [CrossRef]
50. P. F. Devor, N. Tian, I. C. Nishimura, E. M. C. Teng, S. N. Hillman, I. Narayanan, D. A. Ulbert, D. Boas, A. M. Kleinfeld, and Dale, “Suppressed neuronal activity and concurrent arteriolar vasoconstriction may explain negative blood oxygenation level-dependent signal,” J. Neurosci. 27(16), 4452–4459 (2007). [CrossRef]
51. P. A. Duan, B. V. Bedggood, A. B. Bui, and Metha, “Evidence of flicker-induced functional hyperaemia in the smallest vessels of the human retinal blood supply,” PLoS One 11(9), e0162621 (2016). [CrossRef]
52. P. S. Silva, J. D. Cavallerano, J. K. Sun, A. Z. Soliman, L. M. Aiello, and L. P. Aiello, “Peripheral lesions identified by mydriatic ultrawide field imaging: distribution and potential impact on diabetic retinopathy severity,” Ophthalmology 120(12), 2587–2595 (2013). [CrossRef]
53. M. Camino, S. S. Zhang, T. S. Gao, U. Hwang, D. J. Sharma, D. Wilson, Y. L. Huang, and Jia, “Evaluation of artifact reduction in optical coherence tomography angiography with real-time tracking and motion correction technology,” Biomed. Opt. Express 7(10), 3905–3915 (2016). [CrossRef]